What type of control chart...

[Raddog]

should be used for a true position feature of .002. An IX-MR is currently being used at our facility but this doesn't sound right to me. The goal would be to always maintain the true position as low as possible, however it seems the IX chart would require something in the middle.

What would anyone suggest for this type of feature?

 

[D.Scott]

IMHO the IX-MR seems to be right. Your true position recorded as +/- should give you tolerance limits and the true position (0) would then be, as you say, in the middle.

Then of course, maybe I am not seeing the right problem.

Dave

 

[Dave Strouse]

Control charts don't control

In reading your post you seem to be implying that the proper control chart will help you keep the true position at a minimum?
But, control charts only tell you if there is reason to believe the process has changed. They tell you if the true position for any one part is within the range expected from previous data.
I may be assuming, but I suspect that you like many others want to adjust the process based on the current reading or readings. I hope this is not the case, but if it is not I don't understand your comment about wanting a control chart that will maintain to the goal on the low end. The process will maintain to where it will maintain, regardless of the tool to monitor it. No control chart can change that, only process corrective action and improvement work.
As for the type to use, the golden rule is to capture all variation you want to call common cause within the subgroups and all the special causes between subgroups. This is called rational subgrouping. Since I think you are mostly interested in flagging abnormal true position, you would hide those to some extent by any grouping. It has to be an individual chart.

 

[Raddog]

Thanks for your replies. I guess my thinking was that I should use two IX-MR charts. One to monitor the X-axis and one to monitor the Y-axis, this way I would be able to isolate if one of the axis was unstable versus grouping the two axis together. not sure if that's what I want to accomplish, but it sounds good.

As far as subgroups, when monitoring this true position, the part only has this one true position feature and I get one part every 3 hours so my subgroup is one. Is this still the right chart to use. Again, thanks for all your help.